NBER WORKING PAPER SERIES AN ECONOMIC HISTORY OF FERTILITY IN THE U.S ...

NBER WORKING PAPER SERIES

AN ECONOMIC HISTORY OF FERTILITY IN THE U.S.: 1826-1960

Larry E. Jones Michele Tertilt Working Paper 12796

NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge, MA 02138 December 2006

We would like to thank Ran Abramitzky, George Alter, Maristella Botticini, Gregory Clark, Lisa Cook, Marianne Hinds, Ellen McGrattan, and Petra Moser for helpful comments. We would also like to thank seminar participants at Stanford University and Northwestern University for their comments. Several research assistants have helped with this project at different stages: Alice Schoonbroodt, Todd Schoellman, Soohyung Lee, Alejandrina Salcedo-Cisneros, and Adrienne Lin. Financial support from NSF grants No. 0519324 and 0452473 is gratefully acknowledged. All remaining errors are ours. The views expressed herein are those of the author(s) and do not necessarily reflect the views of the National Bureau of Economic Research. ? 2006 by Larry E. Jones and Michele Tertilt. All rights reserved. Short sections of text, not to exceed two paragraphs, may be quoted without explicit permission provided that full credit, including ? notice, is given to the source.

An Economic History of Fertility in the U.S.: 1826-1960 Larry E. Jones and Michele Tertilt NBER Working Paper No. 12796 December 2006 JEL No. J1,J11,N30

ABSTRACT

In this paper, we use data from the US census to document the history of the relationship between fertility choice and key economic indicators at the individual level for women born between 1826 and 1960. We find that this data suggests several new facts that should be useful for researchers trying to model fertility. (1) The reduction in fertility known as the Demographic Transition (or the Fertility Transition) seems to be much sharper based on cohort fertility measures compared to usual measures like Total Fertility Rate; (2) The baby boom was not quite as large as is suggested by some previous work; (3) We find a strong negative relationship between income and fertility for all cohorts and estimate an overall income elasticity of about -0.38 for the period; (4) We also find systematic deviations from a time invariant, isoelastic, relationship between income and fertility. The most interesting of these is an increase in the income elasticity of demand for children for the 1876-1880 to 1906-1910 birth cohorts. This implies an increased spread in fertility by income which was followed by a dramatic compression.

Larry E. Jones Department of Economics University of Minnesota 1035 Heller Hall Minneapolis, MN 55455 and NBER lej@econ.umn.edu

Michele Tertilt Department of Economics University of Pennsylvania 160 McNeil Building 3718 Locust Walk Philadelphia, PA 19104-6297 tertilt@stanford.edu

1 Introduction

If children were a normal good, one would expect richer people to have more children than poorer people. This was precisely the concern expressed by Malthus (1798) who argued that economic growth would lead people to reproduce at a faster rate, and hence food supply per capita would eventually decline. However, while plausible theoretically, empirically it is hard to find such a positive link. On a country level, most countries have experienced a decline in fertility while incomes were rapidly growing. Cross-country data also show a clear negative link between GDP and fertility. In this paper, we take a closer look at this relationship in micro data from the United States. Using census data from nine different censuses we analyze the time series and cross-section dimension simultaneously. We explore the exact nature of the relationship between income and fertility for five-year birth cohorts of women between 1826-30 and 1956-60 and find that it is negative for each cohort. We document how `steep' this relationship is, and find that overall it has been fairly stable. We also conduct several accounting exercises and find that in an accounting sense, increasing incomes can explain up to 90% of the decline in fertility over this time horizon. We also identify systematic deviations from a stable, time-invariant relationship between income and fertility. For example, fertility differentials across income levels first widened and then compressed significantly: fertility was more sensitive to income for women born between 1875 and 1915 ? a period covering both the Fertility Transition and the Baby Bust of the 1930s ? than either before or after.

Facts such as these are just some of the examples of demographic changes that have been seen in the US and in other developed countries over the last 200 years that pose a challenge to modern researchers to explain and understand. Other examples include changes in the incidence and timing of marriage, the increase in female labor supply, changes in the care of parents in old age, an increasing divorce rate, changes in the timing of births, and large increases in the investment in children through education for example. Traditionally, attempts at understanding the causes of these changes and their interrelationships have been conducted in Sociology, History and Demography. They have increasingly become targets for researchers using the standard techniques of Economics, however. Becker (1960), Becker and Lewis (1973), Schultz (1973) and Willis (1973) are early examples with static models while Barro and Becker (1989), Becker and Barro (1988), Galor and Weil (1996, 2000), Alvarez (1999), Fernandez-Villaverde (2001), Greenwood and Seshadri (2002), Boldrin and Jones (2002), De la Croix and Doepke (2003), Greenwood, Guner, and Knowles (2003), Doekpke (2004), Greenwood, Seshadri, and

2

Yorukoglu (2004), Boldrin, DeNardi and Jones (2005), Doepke (2005), Falcao and Soares (2005), Greenwood, Seshadri, and Vandenbroucke (2005), Tertilt (2005), and Boldrin, Jones, and Schoonbroodt (2006) and are examples of more recent, explicitly dynamic analyses. This is quite natural in that many of these decisions are intrinsically dynamic and hence, the development over the last 25 years of the techniques of modern capital theory are particularly useful in understanding the trends seen.

With this recent theoretical literature in mind, our approach is based on trying to identify and present facts that will be useful in the challenge of modelling fertility decisions using economic techniques. Thus, we will, whenever possible, use variables that are closely linked to individual decisions of optimizing households given the information that they have. Our ability to do this is restricted, as always, by limitations in the data. For example, only fertility outcomes are available, not planned fertility. Similarly, for most years, only gross fertility is available not surviving fertility (i.e., net of infant and child mortality). Despite these limitations, the data are of considerable interest.

Beginning in 1900 the U.S. Census contains a question asking women how many children they had over their life - Children Ever Born (CEB).1 This data was collected until 1990 with the exception of 1920 and 1930. Since the 1900 census data has information on the age of a woman, we can use this data to go back to the 1826 birth cohort (by focusing on 74 year old women) to form a long time series of estimated fertility for surviving women. Since the Census data is individual record data, it also contains detailed information on the characteristics of women, their husbands, and other family members that are of use in disentangling fertility patterns over the last 150 years. In particular, the Census contains data on occupation, education, race, and geography. Using Occupation and Education as proxies for income or wealth,2 we construct a cross-section of the relationship between wealth and fertility for five-year birth cohorts beginning with 1826-30 and ending in 1956-60.3 This allows us both to identify separately the cross-sectional and

1All data presented in this paper is based on a 1% sample of the U.S. Census data, made publicly available at by Ruggles et. al. (2004). Our analysis is based on data from the IPUMS webpage available as of 10/20/2004. This data set is updated regularly, the revision history, is available at:

2Indeed, these may be better measures of the relevant variables determining decisions than, for example, income in a single year.

3Our analysis is based on women born between 1826 and 1960, grouped together in cohorts spanning five birth years each. Data for earlier birth cohorts is also available from the 1900 Census, but since it requires information from women that are in their 80's, sample sizes are correspondingly smaller. Hence, we focus on the period since the 1826 to 1830 birth cohort, which we identify with the mid-point of the period and call the 1828 cohort.

3

time series properties of the relationship between income and fertility and, as a result, document how the time series of fertility has differed for women in different parts of the income distribution.

Some of the facts we describe in this paper are commonly known: the overall decline in fertility and the baby bust and boom of the 1930s through 1960s. To paint a complete picture of the U.S. fertility experience, we include a description of these well-known facts in addition to the many novel insights based on the long time series of cross-sectional fertility data that are provided.

Much of the focus of the paper is on the relationship between measures of wealth and fertility. Our main findings regarding the relationship between fertility and occupational income (OI) ? a proxy for life time income ? are as follows.4

First, we find a strong negative relationship between OI and fertility for each crosssection. Our estimate of the overall income elasticity of fertility is -0.38. Evaluated at the mean income for the entire time period, $15,000, this means that a doubling of income implies a fertility fall from 3.2 to 2.5 children per woman.

Second, we find that the observed fertility pattern is surprisingly consistent with the hypothesis that all of the observed fertility decline is the result of a single stable relationship between income and fertility in conjunction with an outward shift in the income distribution over time. We fit an iso-elastic regression between OI by decile of the population and average CEB within that decile for each birth cohort. Using the estimate obtained in this way, we can decompose the observed changes in fertility into movements along the curve due to the growth of income over time and shifts of the curve itself. Between the 1828 birth cohort and the 1958 one, average income increased by a factor of about 13, and our estimated elasticity of fertility is -0.38. Thus, one would expect fertility to fall by 63%, from 5.6 to 2.1 children, whereas the actual fall was to 1.8. Hence, in this sense, 94% of the observed drop in fertility would have been predicted based on the relationship between income and fertility fitted from the 1828 birth cohort.5

Third, we document systematic deviations from such a stable relationship. These include a pattern of fertility that is too low for women born during the 1875 to 1910 period and too high for women born between 1925 and 1940. Not only did the level of the relationship between fertility and income change, but also its slope. We find that the income elasticity of children changed significantly over time, rising from about -0.33 in

4We postpone the precise definition of OI until Section 4. 5Of course the results of this calculation depend on which cohorts are used. We give more details in Section 4.

4

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download